AI and Universal CO2 Capture
AI becomes more powerful every day, but it will never fix stupid!
Figure 1 – CO2 absorbing sponges organized as a forest of very tall trees. Key elements of cyanobacteria have been repurposed to fix our atmosphere.
Exploiting thermodynamic common sense, AI is developing a universal CO2 capture technology right now. The future of DAC is vertical, passive, and energy efficient.
Thought 1: Developing a Universal CO2 capture technology is essential and inevitable.
More than 30 years have passed since world leaders agreed that atmospheric CO2 was a problem. 2025 is half over, and our civilization still does not have a technology that can cost-effectively remove CO2 from the concentrated streams produced by fossil fuel combustion. It is a tragedy of immense magnitude that these very same technologies are the best promoted solutions for removing dilute CO2 from the air.
Using first principles reasoning, I have spent 10 years studying how living organisms' technologies, strategies, and chemistries manage environmental CO2. I have concluded that life is possible and living organisms thrive, even in low-energy environments, because their technology harvests and refines carbon exergy.
This starkly contrasts with today's best and most heavily promoted CO2 capture technologies. These technologies generally destroy exergy by turning carbon-free renewable electricity, which can do work, into worthless heat. They destroy carbon exergy as if that were the mission, and capturing CO2 was some co-benefit.
Al Gore’s recent statements about DAC represent an honest and accurate assessment about today’s best DAC technologies - "ridiculously expensive and hilariously impractical". The former US Vice President is too polite to be blunt. I’m not! Today’s CO2 capture technologies are “thermodynamic idiocy”. In their current form, DAC technologies exist so people get paid pretending they are solving an important problem. The climate grift is on.
Representing thermodynamic common sense, universal CO2 capture is defined by its ability to harness and refine existing carbon exergy. By not destroying this exergy, energy is unnecessary to reacquire the previously possessed carbon exergy. Universal CO2 gas separation and concentration is fundamentally more energy efficient than today’s exergy destroying CO2 capture technologies.
I have identified several universal CO2 gas separation pathways, and the current generation of AI (May 2025) has confirmed my findings. My Model 5 represents the best pathway to date.
The signature metrics for a Universal CO2 gas separation and concentration technology include;
A 2nd-law efficiency of 40% or better to separate and concentrate CO2 from all CO2 containing gas mixtures.
Arbitrarily scalable, as CO2 gas separation only changes the gas's composition, not its chemistry. Minimal (if any) adverse environmental effects.
Does not depend on heat.
Capable of using waste heat to increase the operational 2nd law efficiency, perhaps as high as 80%.
Can remove CO2 from any gas mixture with consistently high 2nd law efficiency. Like the laws of thermodynamics, Universal CO2 capture technologies will disregard the history of CO2. A plot of efficiency against increasing CO2 concentration will be a flat horizontal line. Current plots decline.
Capable of energy-efficient Partial Pressure Staging, where CO2 is efficiently harvested from a stream at progressively lower concentrations.
Can operate equally well on Earth or off Earth. Universal CO2 capture technology will be the best technology for human habitats in space. SpaceX can only succeed in its off-world ambitions with a Universal CO2 capture technology.
Tunable and energy responsive CO2 concentration in the output. Not every sequestration strategy needs pure CO2. Accelerated Mineral Weathering, mining waste, alkaline materials.
Thought 2: Exergy destruction is a key consideration for energy-intensive activities.
Pumped Hydropower Storage (PHS) is an energy management system that is very energy efficient, having a round-trip efficiency exceeding 80%. PHS works by pumping water from a lower reservoir to a higher reservoir when electricity is cheap, and letting that water produce electricity when electricity is more expensive.
Water in the upper reservoir is potential energy, which can be used to produce electricity. This potential energy contributes to a system's exergy. If the potential energy is diminished without doing work, that system's exergy has been destroyed.
The PHS concept works well as long as the water losses are small. Imagine how well the system would work if it were designed to have a large leak. Electricity pumps water against gravity, but the leak allows the water to flow downwards without turning the turbine.
Figure 2 – Exergy destruction is easy
Water flowing through the leak is water that is not producing electricity; it is exergy being destroyed.
Consider the leak being so large that the system is now 10% efficient. This system would be destroying exergy as if it were the mission. This would be an obvious disaster, and the project would never be constructed or funded. A PHS design with a 10% efficiency would never be the basis for scaling PHS systems worldwide.
Today’s synthetic DAC systems, which are being scaled, aspire to be 10% efficient. The efficiency of flue gas CO2 capture systems is even worse. Exergy destruction by CO2 capture technologies is real and measurable and must be minimized to achieve our 2050 CDR goals.
Jarvis provided a great explanation of this exergy destruction. That essay is here.
AI Explains Exergy Destruction
Thought 3: No Urgency, but plenty of Incompetence, Indifference, Climate Profiteering, and Greenwashing
Over the past three years, I have sought funding to develop universal CO2 capture. More than a dozen funding applications have been submitted, including:
Five to the Biden DOE
Three to Breakthrough Energy Ventures (BEV)
Three to Climate Frontier
Two to the NSF
Two to Milkywire.
One to Microsoft
All applications have been rejected. Blaming the messenger is entirely appropriate, but no one has bothered to ask a single question. No one is curious about the conservation of exergy, thermodynamically optimal CO2 gas separation and concentration, or how concepts from the molecular biology domain can be used to build a CO2 capture machine.
All that talk about “1000 shots on goal” and “All hands on deck call for innovation” is just window dressing.
These organizations have their agendas.
The Biden DOE wasted its opportunity to achieve impact by squandering time and money on rewarding “Friends of the Organization.” They prioritized DEI over the pursuit of innovation. Their accomplishments in achieving their Carbon Removal Moonshot can best be described as incompetent or worse.
The Trump DOE is focused on promoting increased coal consumption and is indifferent to the need to reduce CO2 from coal emissions.
Despite its stated goal of supporting climate innovation, BEV's investments in CO2 capture technologies, driven by its venture capital model, have arguably prioritized financial gain over the development of truly effective CO2 capture solutions. They support technologies that are ultimately insufficient in addressing the climate crisis.
The rest of the organizations are greenwashing the public by claiming their efforts help achieve climate goals. The reality is that they are spending money and wasting time on meaningless accomplishments. And yes, removing 100 million tons of CO2 from the air, over ten to fifteen years, is a worthless gesture when our civilization releases 50 gigatons of greenhouse gases annually.
The climate crisis demands we prioritize long-term impact over short-term optics. Otherwise, future generations will inherit a planet featuring runaway warming and ecological collapse.
Thought 4: AI is accelerating the development of a universal CO2 capture technology.
Stone, Bronze, Iron, Industrial, Computer age…
This is the dawn of the Artificial Intelligence Age, and these tools will significantly accelerate innovation. Organizations with valuable domain knowledge will be wildly productive and successful within their industries. I work with all the leading AIs (May 2025), so I will refer to these tools as the “J.A.R.V.I.S Collective” or just Jarvis.
In early 2022, I acquired a critical mass of understanding of how a nature-inspired CO2 capture technology functioned. The initial models were based on materials that did not exist (but could) in topologies that were difficult to produce. I have spent the ensuing years iterating from that impossible point, progressing towards models that can be built.
I think with my hands by building stuff, but I explore in the digital domain by writing software to model physical phenomena. Developing algorithms adapted from the molecular biology domain is challenging. Jarvis has been increasingly helpful in writing the software for my digital laboratory.
Jarvis provides a detailed analysis for each key chemistry in isolation. This “Unit Testing” helps me create the necessary “Proof of Chemistry” experiments.
I have worked with Jarvis to refine the pathways for Model 5 and create a specification document describing the Model 5 CO2 capture system. All the information is in one document. I then instructed Jarvis to write a software simulator based on this specification. Jarvis produces better software in much less time.
The specification is a living document and will be updated with new ideas and feedback from ongoing efforts. The code generation process will be repeated as necessary and whenever upgraded AI models become available.
AI is the greatest tool in the history of tools, and I am using it to solve a challenging and important problem. Working with Jarvis, I will find the most effective way to implement universal CO2 capture by repeatedly asking one question: “Given this system that achieves these goals, suggest ways to accomplish the same task, but differently”.
Thought 5: Role-playing with Jarvis as the world’s best qualified physicist, most accomplished molecular biologist, committed to solving the CO2 gas separation problem.
My above list of metrics defining a universal CO2 capture technology is a subset of the complete list. The full list, along with a scaffold of prompts and guidance in the form of context, is used to challenge Jarvis to innovate.
Knowing one possible solution to creating a universal, energy-efficient CO2 capture technology makes me curious about what Jarvis can do to create other Universal CO2 capture pathways. I frequently ask Jarvis to develop such pathways. So far, Jarvis is limited by the training data. Jarvis makes decisions based on current implementations, popular but flawed metrics, and technically correct but imperfect reasoning (aka The Conundrum). Jarvis is still “Thinking inside the Box”!
Recently, Elon Musk stated that Grok 3.5 is capable of first principles reasoning, emphasizing electrochemistry. This is fantastic! This accomplishment allows Jarvis to innovate beyond the training data. When this reasoning capability is possible, it's “Game Over” for current CDR investments.
From now on, every AI release will be precisely like getting a box of chocolates; the day will be full of joy, excitement, and discovery.
The Point: AI cannot fix stupid
If the researcher is persistent, Jarvis can transform a marginally good idea into a brilliant concept. While AI is a potent tool, thermodynamics is unyielding. AI cannot violate or change the Laws of Thermodynamics.
I have asserted for some time that existing abiotic (or synthetic) CO2 capture technologies (DAC, point source, Ocean electrochemical, etc.) are based on thermodynamic idiocy caused by poor choice of metrics and implementation timeframes. Calcium looping, as implemented by Heirloom and Carbon Engineering, is fundamentally flawed. Adsorption, as implemented by many organizations, is equally flawed. Aqueous amines used by many to capture CO2 from flue gases are flawed as well.
These pathways are flawed in the most fundamental, thermodynamic terms. They enthusiastically destroy carbon exergy, and this reality cannot be fixed. No matter how capable AI becomes, it cannot fix the poor thermodynamics underpinning today's CO2 capture technologies.
A technology founded on optimal thermodynamic principles will survive the test of time, while technologies based on poor thermodynamic principles will be discarded. AI will accelerate this process.
We move towards 2050 every day. The question our civilization faces is this: “When is the best time to abandon thermodynamic idiocy and embrace thermodynamic common sense?”.
Bonus for the investors: Stay safe and wait and watch and keep watching…
Present and future investments into existing CO2 capture pathways are/will trend towards worthlessness. Smart money will find the greater fool and get out!
I have many essay fragments (unpublished) documenting why today’s popular CO2 capture pathways (flue gas, air, ocean (especially the ocean), land (biomass), and rock) cannot achieve meaningful scale to achieve the world’s 2050 carbon removal goals. With the arrival of a dominating, universal CO2 capture technology owned and operated by large organizations like China and Saudi Aramco, these popular pathways will be mostly worthless. Investors should be mindful of how their investments help these large organizations achieve super-gigaton-scale operations.
It is frequently stated in the prompt engineering arenas that your great idea today is subject to the limitations of the following AI iterations. Once AI escapes the training data, there is no proprietary information. This will be challenging for investors in many fields.
Many investors are always proud to note that they do not invest in ideas. They invest in teams that execute. Consider the future happening now: A capable team is working on yesterday’s best CO2 capture concepts, and a new, more capable AI model is released. Tomorrow’s capable team will be able to work with tomorrow’s dramatically better CO2 capture concepts. While executing perfectly, the first team’s efforts are obsolete.
How will investors adapt as AI-driven innovation accelerates rapidly?
Investors avoid innovation risk at all costs. Their best strategy appears to be to sit safely on the sidelines until AI stops improving, even if that process takes decades. If so, kiss the 2050 CDR goals goodbye.
The simplest explanation for no urgency: We are past the point of no return.
With aggressive support, universal CO2 capture could be widely deployed within 10 years for flue gas capture and super-megaton DAC. But there is no urgency.
It may be inappropriate for me to associate leading organizations with Incompetence, Indifference, Climate Profiteering, and Greenwashing. After all, a better explanation is possible. Occam's razor states that among competing explanations, the simplest one is often the most likely to be true.
Organizations such as the DOE, BEV, Frontier, Microsoft, and others are staffed with the brightest minds. They see much more data than most and have teams analyzing it and forecasting the future. These organizations have no sense of urgency and probably concluded a while back that CO2 removal is a pointless exercise.
Our civilization is not reducing CO2 emissions necessary to achieve meaningful goals.
It is increasingly apparent that the damaging forces we have unleashed on our planet are running away (positive feedbacks).
The increasing heat and ocean acidification are irreversible.
But acknowledging this reality would be depressing. So, in the spirit of “Keep a stiff upper lip”, they put minimal effort towards developing CDR technologies, knowing full well that there is no hope.
Michael Landmeier
President, Biosolar Energy Corporation
Founder, XCaptureCO2
#CDR, “Carbon Removal”, “CO2 Capture”, “CO2 Removal”, #Aramco , #spacex , #adnoc, #energy